Pro-Ject Phono Box DS3 B Review

[aagstn]

I don’t get it, $800 for a phono stage! Is it really that much better? The reason I gravitated to this site is because they call this stuff out.

It seems like balanced is the reason for the cost. I guess if you have the budget for a balanced turntable this isn't that expensive. I'm sticking with my Mani 2 at $150.

 

[Zapper]

This graph characteristic is a classic sign of passive RIAA equalisation causing premature clipping.

Could you explain this point in more detail?
If a pre-amp with passive eq looks like this:

then the clipping at high freq at lower input voltage presumably occurs in IC1, because IC2 receives signal with attenuated high freq. But shouldn't the clipping at IC1 be freq independent, assuming it can pass a full amplitude signal up to 20kHz without distortion?

 

[Michael Fidler]

Could you explain this point in more detail?
If a pre-amp with passive eq looks like this:
View attachment 345934
then the clipping at high freq at lower input voltage presumably occurs in IC1, because IC2 receives signal with attenuated high freq. But shouldn't the clipping at IC1 be freq independent, assuming it can pass a full amplitude signal up to 20kHz without distortion?

I can certainly lend my thoughts to that! The clipping does indeed happen in IC1 in a typical preamplifier. Because the output for a given stylus displacement rises at 6dB per octave with frequency for a moving magnet and moving coil cartridge, the kinds of ticks and pops on an LP produce high-frequency, asymmetrical peaks that can exceed 100mV or so and be easily clipped if IC1 has a typical gain of 30dB.

Usually in this circuit you have 30dB in the first stage, 30dB in the second stage, and 20dB lost in the network C1, C2, R3, and R4.

The clipping is independent of frequency, but the cartridge output isn't. I might have a hunt through my old flat-level cartridge transfers to see if I can find some good waveforms to demonstrate this phenomenon.

EDIT - Also we have to consider the maximum output as well, in this case the output is not frequency independent relative to the input either due to the RIAA equalisation. As the total gain of the system at the top of the audio band will be around 20dB or so, there is no excuse to have the maximum output limited by playing amplify-attenuate-amplify.

 

[Zapper]

I can certainly lend my thoughts to that! The clipping does indeed happen in IC1 in a typical preamplifier. Because the output for a given stylus displacement rises at 6dB per octave with frequency for a moving magnet and moving coil cartridge, the kinds of ticks and pops on an LP produce high-frequency, asymmetrical peaks that can exceed 100mV or so and be easily clipped if IC1 has a typical gain of 30dB.

Usually in this circuit you have 30dB in the first stage, 30dB in the second stage, and 20dB lost in the network C1, C2, R3, and R4.

The clipping is independent of frequency, but the cartridge output isn't. I might have a hunt through my old flat-level cartridge transfers to see if I can find some good waveforms to demonstrate this phenomenon.

Thanks, that makes sense for real usage of the preamp. But what about Amir's test? I assume the X axis represents the actual unequalized voltage from the analyzer. If that's true it is odd that IC1 would clip at much lower generator level at high freq. Or is the recording RIAA eq applied between the generator level output and the pre-amp?

 

[Michael Fidler]

Thanks, that makes sense for real usage of the preamp. But what about Amir's test? I assume the X axis represents the actual unequalized voltage from the analyzer. If that's true it is odd that IC1 would clip at much lower generator level at high freq. Or is the recording RIAA eq applied between the generator level output and the pre-amp?

View attachment 345939

It will either be pre-equalised, or set to generate a reference constant level at the output, as is typical for such situations.

If it's driven to clipping, it's likely to be a RIAA script matched to constant output, rather than an output tracker as the gain will change as the circuit goes nonlinear in overload.

 

[DSJR]

Has anyone shown that high fidelity reproduction of pops and clicks, with all their high frequency energy, is audibly superior to clipping them at a lower amplitude? Passing a 100mV click @ 40dB gain is a 10V output from the pre-amp - that's going to be bloody loud if the power amp can pass it along unclipped to the speakers. 26dB louder than the recorded material at 5mV level. And there is a good chance of clipping in the power amp instead, depending on the listening level.

It's not a-priori obvious to me that a click of that magnitude wouldn't sound better if clipped to a lower amplitude, say 5V.

Back in the 90's, AVI made a preamp with a built-in phono stage (S2000MP+P) which was designed for a high headroom before overload, but accepting that some pickups have massive peaks on ticks and splats, the overload was designed to SOFT clip over I believe 70mV which the designer favoured. In practise and after the 45 minutes it *always* took for the stage to 'open up' subjectively (many samples all the same), it was superb and the *effects* of surface noise were never ever intrusive with a variety of cartridges, mostly MM types.

This later version (post 1996 or so) preamp was well tech-reviewed in HiFi News at the time but I can't now remember if the phono stage was tested too as it was an extra which one factory- ordered to be fitted at the time before being shipped out. The board was solidly encased in a metal cover as was the power supply and I never took mine apart to take a look sadly.

 

[Michael Fidler]

So I've done a little digging through my transfer library to find a suitable example of what a flat level transfer of an LP with not too major surface damage/scratches looks like. The top waveform shows the 'flat' (no RIAA) transfer of a mono disc summed to mono (for ease of viewing - if you include vertical elements it gets worse) and shows how unequalised ticks and pops from the cartridge are exceeding the recorded level (a fairly hot first track) by 20dB or so with ease, hence the need for lots of HF headroom in a phono preamp.

The trace at the bottom shows the RIAA version with the eq applied in software. Notice how small the ticks and pops are now, some are quite invisible. The frequency content of these is mainly in the 10-20kHz region and is thus attenuated 5 to 10 times by the RIAA curve. If they are clipped/overloaded before the RIAA curve, then this will generate lower frequency distortion products outside the attenuated 10-20kHz zone. These will not be attenuated by the RIAA curve and be more audible than the high frequency tick itself. Not nice!

 

[restorer-john]

So I've done a little digging through my transfer library to find a suitable example of what a flat level transfer of an LP with not too major surface damage/scratches looks like. The top waveform shows the 'flat' (no RIAA) transfer of a mono disc summed to mono (for ease of viewing - if you include vertical elements it gets worse) and shows how unequalised ticks and pops from the cartridge are exceeding the recorded level (a fairly hot first track) by 20dB or so with ease, hence the need for lots of HF headroom in a phono preamp.

View attachment 346007

The trace at the bottom shows the RIAA version with the eq applied in software. Notice how small the ticks and pops are now, some are quite invisible. The frequency content of these is mainly in the 10-20kHz region and is thus attenuated 5 to 10 times by the RIAA curve. If they are clipped/overloaded before the RIAA curve, then this will generate lower frequency distortion products outside the attenuated 10-20kHz zone. These will not be attenuated by the RIAA curve and be more audible than the high frequency tick itself. Not nice!

Again, all this goes to show is a high overload margin is not only desirable, but essential for quality phono reproduction. But this is also something we've known for 50+ years.

Most of the good phono stages through the 70s and 80s ran high voltage front ends to give MM overloads of 200-300mV (at 1kHz)- figures I have confirmed with plenty of my vintage front ends and figures you just don't see anymore due to every man and their dog 'building' 'phono stages' with OPAMPs instead of discrete components running from decent rails.

When did you last see specifications like this? (40 years ago).

The Project clearly uses a combination of passive and active RIAA- hence the pathetic overload characteristics that decrease massively at HF.

But @Michael Fidler I think this discussion should be hived off into a phono topology thread especially now, as you, a competing phono stage manufacturer, have weighed in.

 

[AaronJ]

Again, all this goes to show is a high overload margin is not only desirable, but essential for quality phono reproduction. But this is also something we've known for 50+ years.

Most of the good phono stages through the 70s and 80s ran high voltage front ends to give MM overloads of 200-300mV (at 1kHz)- figures I have confirmed with plenty of my vintage front ends and figures you just don't see anymore due to every man and their dog 'building' 'phono stages' with OPAMPs instead of discrete components running from decent rails.

When did you last see specifications like this? (40 years ago).

View attachment 346030



The Project clearly uses a combination of passive and active RIAA- hence the pathetic overload characteristics that decrease massively at HF.

But @Michael Fidler I think this discussion should be hived off into a phono topology thread especially now, as you, a competing phono stage manufacturer, have weighed in.

I have wondered, what exactly does input level 2.5mV mean?

 

[restorer-john]

I have wondered, what exactly does input level 2.5mV mean?

2.5mV was the defacto rated standard 'output' of a typical MM cartridge at 1kHz 5cm/S and phono stages sensibly rated their sensitivity to match.

3.5-5.0mV would have been a 'hot' cartridge with DJ cartridges sometimes rated even higher.

 

[AaronJ]

2.5mV was the defacto rated standard 'output' of a typical MM cartridge at 1kHz 5cm/S and phono stages sensibly rated their sensitivity to match.

3.5-5.0mV would have been a 'hot' cartridge with DJ cartridges sometimes rated even higher.

So is that basically indicating that 2.5mV would be enough to drive the amp to maximum output into 8 ohms?

 

[Endibol]

You're begging the question - you are simply assuming the answer, which is that clipping is bad. I'm asking for evidence that a clipped pop or click signal of reduced magnitude sounds worse than a non-clipped signal of enormous 26dB+ magnitude (assuming the pre-amp overload recovery is fast and stable).

That's assuming that the rest of the chain has 26dB+ of headroom and won't clip either. If not, then clipping occurs in the power amp, which is likely worse than the pre-amp.

A clipped signal will have a lot of high frequency content that is more disturbing to our ears.

 

[restorer-john]

So is that basically indicating that 2.5mV would be enough to drive the amp to maximum output into 8 ohms?

Typically yes, back when sensitivities were normally 150mV for line levels for full rated power or output.

A phono stage at 2.5mV for 150mV output is 35.6dB gain, which is right on the average for RIAA stages in amps and preamps made through the 70s-90s.

Line inputs were also typically 150mV for rated output, which meant of course, a 2.0V CD input would result in most amplifiers having their volume control in the first 130 degrees of travel.

 

[restorer-john]

Or is the recording RIAA eq applied between the generator level output and the pre-amp?

It's an inverse RIAA applied to the generator signal output.

If that's true it is odd that IC1 would clip at much lower generator level at high freq.

It's not odd at all. If you have a front end gain stage, followed by the passive RIAA, then another gain stage, the front end will obviously be more prone to overloading at high frequencies as the input to that stage is unequalized at that point and up to 20dB above the reference level at the top end. This Project unit is a combination passive/active stage.

 

[dlaloum]

Amir, thank you for the review!

However, Loading pF, Loading Ohms, Subsonic, 10R, 30R,...,1k - embarrassed to say I don't understand what any of these buttons do

Especially for MM (high output, high impedance) cartridges, the loading, both capacitive and resistive, creates an EQ filter, which is required to achieve a proper flat frequency response.

Different cartridge and stylus combos, require different loadings to adjust them... although 47k became the resistive load "standard" - many of the manuals for cartridges from the 70's state things like "load of up to 75kOhm", and optimal capacitance would range from 100pf to 500pf for different cartridges...

Also, when setting up a cartridge with a non-original stylus, the resulting combination is highly likely to require a differing EQ to achieve a flat F/R...

A phono stage with fully adjustable C and R loading is essential for anyone who really wants to optimise the performance of their cartridge/stylus.

MC cartridge have different behaviour, they are pretty much insensitive to capacitance variations, and have a minor variation to resistive loading....

Regardless of this phono stages flaws in terms of headroom, the loading convenience is an important positive feature... I wish this type of setup were standard on phono stages!

 

[solderdude]

Too bad the overload characteristics (as John mentioned) aren't visualized and the ticks in Michael's file aren't shown in sample detail.
But this would be great in a separate thread.

How 'nasty' clipping becomes is highly dependent on the used components (opamps) and whether or not clipping occurs by limiting the input signal before it goes into the gain stage or that it is the opamp running out of output voltage and saturating and thus how it recovers.

I like the balanced feature which could help battling magnetic induced hum in certain cases. Love the adjustability options.
My own design pre-amp, also hybrid, also has several loading options including 22k and 100k and even a 1k series load option.

The 'problem' is that you either need specs from the cartridge manufacturer (about recommended loading it which often is not available) or need to know what you are doing using a test record when fiddling with capacitance settings. One can do it by ear but might not end up with the actual needed load.
Highly cartridge dependent too.

 

[DrCWO]

Even better is a rare feature: balanced input and output.

Because of this I voted Great

 

[DrCWO]

Unfortunately you're quite correct. It's very difficult to exclude hum when measuring some 60dB at 50Hz with an analyser...

This is why I vote for connecting all via XLR, also the cartridge. The cartridge is a symmetric device by nature!

Doing so you just have to make sure that non of the outputs of the cartridge is connected to its housing like Shure did. But this can be removed and the housing of the cartridge must be connected to XLR ground.

This fixed all my hum issues In the past.

 

[ban25]

Not sure this can be compared to Schiit Mani because it’s a different beast. But can it justify costing twice the Schiit Skoll? Absolutely not. I considered one of these for a little while and have generally felt that Project gear is marked up around an additional 100% from where it should be. They make 39 different turntables not including the dozen or so special editions. Feels like IKEA of the turntable world.

Not to shit on the Skoll , but there are multiple owners reporting an audible buzz in one channel. Probably sample variation, but if I had one with that issue, I'd have to send it back.

 

[trinitronx]

As the cartridge is a floating source, there is no possibility of creating a ground loop unless the turntable wiring is faulty

Which is unfortunately rather common with certain entry-level (a.k.a. "cheapo") turntables. For example, the original AT-LP120-USB phono cartridge signal path has a non-bypassable L-C filter circuit always in place when set to "Phono" mode (e.g. when supposed to bypass the internal preamp). Additionally, there appears to be a SMD component jumpering all of the cartridge's negative lines as they come out from the tonearm and into the circuitboard inside the chassis. Audio-Technica was so "kind" as to include this pre-soldered ground loop path which is permanently installed for the user.

Another example would be this Udreamer 009 turntable which is currently popular on Amazon. Testing this one revealed multiple issues, including a built-in ground loop when in bypass mode. Another hilariously bad issue with this one: The "Line" / "Phono" switch was actually labeled in reverse! So, when switched to "Phono", actually a Line-level signal was output on the RCA jacks, except without any bass, and with very harsh high frequency range. It seemed like they forgot to apply the RIAA curve when in internal preamp (Line-level) mode (even if it was labeled backwards). There was no continuity between RCA pin and shield with the headshell pins when the switch was set to "Phono" mode and the level of the signal was at Line-level instead. When switched to "Line" mode, as labeled, then the headshell pins were connected to the RCA jacks (as it would be for Phono / "bypass" mode). However, both negative (L-, R-) lines were shorted to each other AND the chassis ground lug. So yet another built-in ground loop path when chassis lug & an external preamp ground lug are connected.

A final example, is even mentioned on the Pro-ject site regarding the use of certain MM cartridges:

Why should I avoid using a moving magnet cartridge?

Moving magnet cartridges are also inherently balanced transducers but the balanced signal ends at the cartridge output pins. For noise suppression in unbalanced systems, manufacturers usually tie the right channel grounding conductor to the cartridge’s chassis. This renders the signal unbalanced henceforth and consequently, balanced connection post-cartridge will introduce grounding issues that’ll generate more noise, but just in the unbalanced right channel. Here’s that graph again to show you just what we mean (hone in on Graph 2!)…

Graph 1: typical moving magnet (MM) phono cartridge connected with standard RCA/ground interconnects (noise identical in both channels)
Graph 2: typical MM cartridge connected with Mini XLR or XLR interconnects (higher noise level of -55dB in right channel vs. -61dB on left channel – not advisable!)
Graph 3: typical moving coil (MC) phono cartridge connected with standard RCA/ground interconnects (significantly quieter than MM but significantly noisier than balanced as shown in Graph 4)
Graph 4: typical MC cartridge connect

This issue may not affect all moving magnet type cartridges, and some include a removable ground clip (usually on the R- line, but sometimes L-). Shure has this page describing how to remove the jumper clip for their cartridges.

So, as with most things of sufficient complexity... "The devil's in the details"... and there's a lot of details to get wrong! (Also, the "great" thing about standards is there are so many to choose from ) It seems Pro-ject is recommending to use MC cartridges as a guaranteed way to achieve a truly floating & balanced signal path from the headshell. Presumably this is because the moving coil design is less likely to allow for manufacturers of the cartridge to short one of the - signal pins.